Fuzzy predictive modelling using hierarchical clustering and multi-objective optimisation for mechan

In this paper, a systematic data-driven fuzzy predictive modelling mechanism is proposed. In this mechanism, a hierarchical clustering algorithm, the agglomerative complete-link clustering algorithm, is first employed to extract an initial fuzzy model. Then, the initial model is improved and optimised through a two-step operation using evolutionary algorithms (EAs), which takes into account both the accuracy and the interpretability performance of the fuzzy system. The proposed modelling approach is tested on a high dimensional problem using real data from the steel industry which concerns the prediction of the mechanical properties of alloy steels, in particular the Tensile Strength (TS). Experimental results show that the proposed approach is effective in eliciting accurate and interpretable models while ensuring their future easy maintenance.